Proceedings of GT 2006
ASME Turbo Expo 2006: Power for Land, Sea & Air
May 8-11, 2006, Barcelona, Spain
GT2006-90270
FLYWHEEL ENERGY STORAGE SYSTEM FOR NAVAL APPLICATIONS
Co Huynh, Patrick McMullen, Alexei Filatov, Shamim Imani
Calnetix Inc., 12880 Moore Street, Cerritos, CA 90703, USA
Hamid A. Toliyat, Salman Talebi
Department of Electrical Engineering, Texas A&M University, 3128 TAMU, College Station, TX 77843, USA
ABSTRACT
A recent trend in designing naval ships is to improve
performance through using more electric equipment. The
reliability and quality of the onboard electric power, therefore,
becomes critical as the ship functionality would entirely depend
on its availability. This paper investigates the possibility of
using Flywheel Energy Storage Systems (FESS), similar to
those earlier developed for commercial applications, to address
issues related to onboard power supplies. A design of a FESS
for onboard power backup and railroad electrical stations is
presented. The FESSs power output parameters are
500kW×30sec in high-duty mode and up to 2MW in pulse
mode. High power output is one of the main advantages of
FESS over commercially available electrochemical batteries.
The other advantages include essentially an unlimited number
of charge/discharge cycles, observable state of charge and
environmental friendliness. Designs of the main FESS
components are discussed: low-loss magnetic bearings, an
energy-storage hub, a high-efficiency motor/generator and
power electronics.
Smaller diesel gensets may additionally be used as power
backups for certain critical loads, such as computer equipment.
There are many unique challenges that must be considered
when developing on-board electrical power supply networks
[3]. Electrical equipment on a ship will include loads that
require very high power for short periods of time, which can be
in the order of a few seconds or milliseconds. Examples
include electromagnetic aircraft launch systems and highenergy weapons. Another consideration is the wide variety of
electrical loads that range from propulsion engines to highly
sensitive computer equipment. The uninterruptible power
supply for the latter is of special concern as computers control
all vital functions of a modern ship. An extremely important
requirement for a shipboard electrical network is high
survivability. Should the ship experience some damage in a
battle, it cannot become completely paralyzed because of the
loss of its primary electrical power supplies. Several leading
defense contractors are currently involved in development of a
distributed onboard electrical power network which would
address this issue. Finally, the equipment installed on a ship
needs to be as compact and power-dense as possible as cargo
space is a premium onboard ships.
INTRODUCTION
Similar to the aerospace industry with its more electric
aircraft initiative to improve the overall system performance,
military vessels are moving towards more electric based
equipment. Gas turbines in combination with electrical
generators are expected to be the main sources of the electrical
power for the new ships [1,2]. Their advantages over diesel
generators include higher power to weight ratio, very low
maintenance, and higher frequencies of acoustic noises, which
are better absorbed by water making a ship harder to detect.
Most of these issues are not exactly new and specific to
naval applications. They are quite often encountered in modern
commercial electrical installations. For example, coexistence of
high-power machinery and sophisticated computer equipment
is very common these days in such applications as aircrafts,
modern trains and cars. Another example is in manufacturing
plants where process machinery demands high power during
initialization and grids need to be stiffened with supplemental
power sources.
1
Copyright © 2006 by ASME
These and other power quality issues are addressed by
using Uninterruptible Power Supplies, Power Conditioning,
Load Leveling and Energy Storage Systems. Traditionally,
these are based around electrochemical batteries. Commercially
available batteries, however, are not capable of high
charge/discharge rates, have limited cycle life, unobservable
state of charge, require high maintenance and impose disposal
problems. Aggressive chemicals used in batteries, such as
acids, are especially dangerous for naval applications where
they can be released if the ship is damaged in a battle.
FESSs provide an attractive alternative to electrochemical
batteries, free of all of the above mentioned disadvantages. A
FESS stores energy in the form of kinetic energy in a rotor
spinning at a high-speed in a vacuum. In a modern FESS, the
rotor is supported in non-contact magnetic bearings offering
essentially unlimited lifetime and charge/discharge cycles.
Recent advances in permanent magnet motor/generator
technology resulted in flywheel charge/discharge rates many
times higher than those achievable with electrochemical
batteries.
In the commercial realm, a new and very promising
application for FESSs is in load leveling and energy recovering
at electrical train stations. Electrical trains consume extremely
high power for short periods of time during acceleration,
dissipating this power in the form of heat later when the train
brakes. The current approach to this problem is sizing the
electrical grid for the peak power rather than the average.
Besides being expensive by itself, this solution implies higher
operational cost in the future since customers pay a premium
for peak power consumption. In addition, existing electrical
infrastructure is both difficult and expensive to upgrade to
accommodate larger and/or more frequent trains required to
address growing commuter needs. FESSs installed at the train
stations allow accumulating the energy released during the train
brake in order to reuse it later for the train acceleration. In
contrast to conventional electrochemical batteries, a FESS is
capable of sustaining a large number of charge/discharge cycles
required in this application for many years without degradation.
a much smaller flywheel hub and a motor/generator than
designs relying on ball bearings. In spite of higher speeds, the
life of magnetic bearings was essentially unlimited due to the
non-contact nature of the suspension. On the other hand, the
flywheel hub design was rather conservative as it was made of
high strength 4340 steel rather than more exotic composite
materials [11]. Using the 4340 steel avoids issues associated
with integration of composites which results in a robust and
predictable design.
One of the most important issues related to application of
the flywheel onboard a ship is dynamic loading. Whereas a
composite rotor offers an advantage of smaller dynamic inertia
forces due to a lower rotor weight, the magnetic suspension
system described in this paper takes advantage of the magnetic
properties of a steel rotor to passively lift it using permanent
magnets. The entire load capacity of the active magnetic
bearings in this case is used to compensate for dynamic
loadings. Such a passive magnetic lift would be very difficult to
realize with a composite rotor due to the non-magnetic nature
of the composite materials.
Table 1 below summarizes the requirements for energy
storage in several military and commercial applications,
comparing what exists today and what is needed for new
applications. A 1.7-4.2kWh 500kW high-duty-cycle / 2MW
low-duty-cycle FESS described in this paper targets both Navy
onboard power backups and commercial railroad applications.
In onboard power backup, the FESS works together with a
diesel genset, providing power for critical equipment in case of
a failure of the main gas turbine or the main electrical network.
Higher power output and energy storage capacity values can be
obtained by connecting several FESS units in parallel.
The starting point for the flywheel described in this paper
was a 125kW×16sec flywheel developed earlier for commercial
UPS applications [4-10]. This design has been shown to offer
an advantageous compromise between novelty and
conservatism. On one hand, it utilized state-of-the-art
technologies such as patented non-contact magnetic bearings
and an advanced permanent-magnet motor/generator. Higher
rotational speeds due to the use of magnetic bearings resulted in
Existing systems
Power
Duration
Stored energy
Recharge time
System output voltage, DC*
System output current, DC*
Vycon 125 kW
flywheel DC source
125kW
17 sec
0.6 kWh
Variable
400-900V
320A
2MW
generator
2MW
continuous
N/A
N/A
800V
2.5kA
Aircraft
Launch
120MW
2 sec
34kWh
45 sec
1.5-4kV
18-50kA
NAVY applications
New commercial applications
High-Energy
Weaponry
50GW
6-10msec
28kWh
6sec
10kV (pulse)
10MA (pulse)
Sacramento
New York
light railroad
heavy railroad
500kW
2.5MW
25.8 sec
12.6 sec
3.58 kWh
8.75 kW h
18 sec
10 sec
600-1,000V (typical)
840A max
4.2kA max
Onboard
Power Backups
1-8MW
0.5-15sec
4.2kWh
1000V
1kA max
* DC bus values are used for comparison between applications
Table 1. Parameters of the existing Calnetix/Vycon systems and requirements for Navy and commercial applications
for FESS.
2
Copyright © 2006 by ASME
NOMENCLATURE
Fz
K r, K z
Id, Iq
Is
iA, iB, iC
iR, iS, iT
Pout
uA, uB, uC
uAB, uBC, uAC
uR, uS, uT
uRS, uST, uRT
νA, νB, νC
νAB, νBC, νAC
νR, νS, νT
νRS, νST, νRT
δ
ωr
θ
axial force exerted on the rotor by the passive
lifter
radial and axial rotor suspension stiffnesses due
to the passive lifter
RMS (Root-Mean-Square) values of the direct
and
quadrature
components
of
the
motor/generator control currents
RMS value of the motor/generator phase current
phase currents on the grid side of the power
electronic unit
phase currents on the motor/generator side of the
power electronic unit
Output power of the generator
phase to neutral voltages on the grid side of the
power electronic unit
phase to phase voltages on the grid side of the
power electronic unit
phase to neutral voltages on the motor/generator
side of the power electronic unit
phase to phase voltages on the motor/generator
side of the power electronic unit
phase to neutral EMFs (Electromotive Force) on
the grid side of the power electronic unit
phase to phase EMFs on the grid side of the
power electronic unit
phase to neutral EMFs on the motor/generator
side of the power electronic unit
phase to phase EMFs on the motor/generator
side of the power electronic unit
torque angle in vector motor / generator control
circular rotational frequency of the rotor
(rad/sec)
angular position of the rotor (rad)
FLYWHEEL TOPOLOGY
The flywheel layout is schematically shown in Figure 1. It
consists of the following major components:
1) Flywheel hub made of high-strength 4340 steel
2) A system of two homopolar PM (Permanent Magnet) biased magnetic bearings; radial on the top and combo
at the bottom
3) Passive PM lifter
4) PM Surface-Mounted (PMSM) 4-pole motor/generator
Another critical element to the FESS is the power electronics.
While not shown in Figure 1, it will be discussed in more detail
along with the function and operation of each of the major
FESS components below.
FLYWHEEL HUB
Similar to its smaller 125kW predecessor, the proposed
flywheel utilizes a hub made of high-strength 4340 steel. A
possible alternative to 4340 steel would be to use modern
composite materials offering much higher strength in certain
directions combined with much lower density. For example,
Top radial magnetic bearing
Radial load capacity: 1350N
PMSM motor
500kW @ 6500RPM
750kW @ 10000RPM
Passive lifter
Load capacity: 15600N
1500
Flywheel hub
(4340 steel)
Combo magnetic bearing
Radial load capacity: 2700N
Axial load capacity: 5350N
Figure 1. Proposed layout and major dimensions of
500-750kW, 1.7-4.2kWh high-duty cycle FESS (the
same flywheel is expected to be capable of
delivering 2-3MW in pulse mode). All dimensions are
in mm.
T1000GB-12000 composite material manufactured by
TORAYCA [12] has maximal ultimate tensile strength of
6370MPa (924ksi) combined with 1800kg/m3 (0.065 lb/in3)
density, whereas the corresponding values for 4340 steel are
1790MPa (260ksi) and 7800 kg/m3 (0.28 lb/in3) respectively
[13]. There are, however, serious risks associated with using
composite materials. In particular, their integration into the
rotor is significantly complicated by strong anisotropy of the
mechanical characteristics, low elasticity modulus and large
differences between temperature expansion coefficients of the
composite and metallic parts of the rotor.
The main advantage of using the composites in the
application considered in this paper would be lower weight of
the rotor for the same amount of stored energy. The lower
weight would also result in smaller magnetic bearings needed
to support this rotor and smaller bearing power consumption.
However, with a steel rotor an alternative strategy can be
applied to achieve the same result: utilize large magnetically
permeable surface of the steel rotor to passively lift it with
permanent magnets. Active magnetic bearings in this case are
only utilized to stabilize an otherwise unstable system and
compensate for dynamic loads. The dynamic loads due to the
linear accelerations will be higher in the case of a steel rotor
due to higher weight. However, since the active bearings do
not have to deal with the static weight, they can offer more
dynamic load capacity within the same envelope. The details
of the passive lift feature are given below.
An important factor that must be considered when sizing the
flywheel hub is the duty cycle of the flywheel operation. This
factor varies greatly between the applications for power
backup, Electromagnetic Aircraft Launch (EMALS), high
energy weaponry and commercial railroad.
3
Copyright © 2006 by ASME
In power backup applications, the flywheel stays in a fully
charged state most of the time, with the rotor spinning at a
constant speed and subjected to constant centrifugal stresses.
The flywheel is discharged only during power supply failures
which are not expected to occur on a regular basis.
stresses caused by rotor unbalance. The samples were failing
after 450,000 cycles, indicating significant safety margins for
UPS applications with a cycle life requirement of 2,000 cycles.
In contrast, when using railroad applications for comparison,
the flywheel is charged and discharged many times during a
day whenever a train comes into a station and leaves it. Since
the life of the flywheel rotor is limited mainly by the fatigue
issues during stress cycling, the same flywheel will last much
longer when used in power backup than in the railroad
applications.
A unique feature of the 500kW flywheel design is a passive
permanent magnet lifter which effectively supports the entire
weight of the flywheel rotor (approximately 15600N or
3,500lb). The steel hub serves as a part of the magnetic circuit
producing the lift force. This feature would be much more
difficult to implement in the case of a composite rotor since
composite materials are magnetically neutral.
Because requirements for the energy storage capacity in two
applications were rather close, it was tempting to develop a
single flywheel design that could be used in both cases. The
above mentioned differences between the applications could be
addressed by using different operating speeds in order to store
more energy with smaller cycle life in power backup, and
ensure longer cycle life in railroad systems by sacrificing some
of the energy storage capacity.
While this lifter can support the weight of the entire rotor, by
itself it is not sufficient to realize a stable non-contact
suspension of the rotor in all directions. This is because, as
with any magnetostatic system, the proposed passive lifter
obeys a very old physics theorem attributed to Earnshaw [14]
with a later expansion by Braunbek [15,16]. In a simple form
tailored to magnetic suspension, this theorem states that the
sum of suspension stiffnesses in all directions in any static
system composed entirely of permanent magnets and softmagnetic components is always less than or equal to zero. In
particular, for an axisymmetric system such as shown in Figure
1, the Earnshaw’s theorem is reduced to the following equation:
Table 2 gives a few examples of possible combinations of the
system parameters resulting from operation at different speeds.
Higher speeds allow for higher energy storage capacities but
result in fewer cycles to failure. Note that at this point the
parameters are calculated without taking advantage of advanced
power electronics being developed at Texas A&M University,
which would allow boosting the output power of the system by
approximately a factor of four for short periods of time. All
parameters in Table 2 are calculated for the same flywheel
design shown in Figure 1.
Preliminary estimates of the rotor cycle life shown in Table 2
are based on the analysis conducted on the existing 0.64kWh
125kW product. This configuration has gone through extensive
fracture mechanics analysis and fatigue sample testing. Possible
material voids and their imparted limitation on the rotor life
have been taken into account during the analysis. Fatigue
samples have been exposed to stresses varying in time from the
maximum value which would be encountered in the flywheel
rotor under real operating conditions to zero and back. With
regards to the maximum stress value, it should be noted that the
Adaptive Open Loop Cancellation feature [6] implemented in
the active magnetic bearing supporting the flywheel rotor
allows it to spin about the center of mass, eliminating additional
Specification
Rated Power Output
Output Duration at Rated Power
Usable Energy
Operational Speed at Full Charge
Maximum Duty Cycle, discharges/hour
Rotor Cycle Life, cycles to failure
Footprint
Weight
PASSIVE LIFTER
K z + 2K r ≤ 0
where K z = −
Kr = −
∂Fr
∂r
∂Fz
∂z
is the axial suspension stiffness, and
is the radial suspension stiffness. Since the
necessary condition for stability is that the suspension stiffness
is positive in all directions, the above equation implies that
stability is not possible. Figure 2 shows the magnetic flux
distribution in one of the possible designs of the passive lifter
producing 15600N of the axial force.
The suspension is stabilized by means of two PM-biased active
magnetic bearings: a combo (axial and radial) bearing at the
bottom of the unit and a radial bearing at the top. In addition to
stabilizing the system, these bearings compensate for
inaccuracies of the passive weight offset and react dynamic
loading. Because of the PM-biased and homopolar
construction, the bearings feature low rotational losses and high
efficiency, which makes them very well suited for flywheel
500 kW
500 kW
12 sec
1.7 kw-hr
8000 RPM
60
1,500,000
1.98 m x 0.85 m
3400 kg
500 kW
500 kW
24 sec
3.4 kw-hr
9000 RPM
30
200,000
1.98 m x 0.85 m
3400 kg
750 kW
750 kW
20 sec
4.2 kw-hr
12000 RPM
4
5,000
1.98 m x 0.85 m
3400 kg
Table 2. Possible combinations of the energy storage system parameters as functions of the
operating speed range. The flywheel rotor and the motor/generator are the same in all cases
(refer to Figure 1.)
4
Copyright © 2006 by ASME
applications.
variation due to the above negative stiffness will be limited to
±2555N (584lbf), less than a half of the axial load capacity of
the combo bearing shown in Figure 1. A more intensely
detailed system design during the next phase of the program is
expected to lead to even lower variation in force due to axial
movement.
ACTIVE MAGNETIC BEARINGS
Figure 2. Magnetic flux in one of the designs of the
passive magnetic lifter producing 15600N (3,500lbf)
of axial force. Sintered 32MGOe NdFeB magnets are
assumed.
While there are plenty of ways to design a passive magnetic
system capable of producing 15600N of force, it is desirable to
do it so that the passive lifter exerts as little stiffness (force
gradient) in all directions as possible. In other words, the force
produced by the lifter should not be strongly dependent upon
the position of the flywheel rotor. This is because the design of
the active bearing control algorithm becomes much more
complicated in the presence of significant negative stiffness and
the performance of the final design is negatively affected by it.
In the present design, the lift force is produced by exposing the
top steel surface of the hub to a magnetic field generated by
stationary permanent magnets. Because the surface is large, the
required force can be obtained with a relatively low magnetic
flux density and large air gaps, resulting in very low values of
the force gradients.
Two active magnetic bearings installed on the opposite
ends of the flywheel rotor are used to stabilize an otherwise
unstable system, compensate for the inaccuracy of the passive
rotor weight compensation and react dynamic loading. The top
bearing provides radial support of the corresponding rotor end
while the bottom bearing has a dual function of radial and axial
support. Because of homopolar PM-biased construction, both
bearings feature high efficiency and low rotational losses. The
design and operation of the top radial active magnetic bearing
are clarified by Figures 4 and 5.
Radial control
windings
Rotor
laminations
Bias magnet
Solid iron
rotor parts
Solid iron
stator poles
Figure 4. Axial section of the Radial Magnetic Bearing
and bias flux flow.
Y
Bias Flux
Figure 3 shows how the axial force produced by this lifter
depends on the axial gap. With the nominal gap (7.6mm or
0.3in), the axial force gradient was calculated to be 5.11⋅106
N/m (29,200 lbf/in). If the axial movements of the rotor are
limited by backup bearings to ±0.5mm (0.02in), the axial force
Control
Flux
FY
X
Passive lifter, Fz vs z
Axial force, N
Stator
laminations
20000
18000
16000
14000
12000
10000
8000
6000
4000
2000
0
Control
Coils
Figure 5. Radial section of the Radial
Magnetic Bearing and explanation of the
radial force generation.
0
2
4
6
8
10
12
Axial gap, mm
Figure 3. Axial force produced by the passive lifter
as a function of the axial gap.
The bias flux in this bearing is generated by axially
magnetized permanent magnets, a solution that results in a
more compact and efficient design than when current-carrying
coils are used for this purpose. The bias flux flow in the bearing
axial plane is illustrated in Figure 4. Currents in the control
5
Copyright © 2006 by ASME
coils produce magnetic flux in the radial plane of the laminated
active pole and the laminated rotor part, which, when
superimposed on the bias flux, result in radial force
proportional to the control current.
Figure 5 clarifies the mechanism of generating a radial
force. In the upper pole, the control and bias fluxes add up,
while in the lower pole they subtract. Higher net flux density in
the upper air gap results in the radial force acting in the positive
Y direction. An advantageous feature of this design is that the
control flux does not flow through high reluctance permanent
magnets. This minimizes size of the control coils as well as
their power dissipation.
The mechanism of generating a radial force in the
combination bearing is very similar to that in the radial bearing.
The distribution of the bias flux in the combo bearing is
illustrated in Figure 6. Within the stator lamination stack flux
flows similar to the radial bearing, and introduction of a radial
control flux results in a radial force (see Figure 5). The
difference between the two bearings is an additional axial
control coil which produces the control flux in the axial plane.
The axial control flux, when superimposed on the bias flux,
results in the axial force. This mechanism is illustrated in
Figure 6. The bias and control fluxes sum at the upper axial
pole and subtract in the lower, resulting in higher flux density
at the upper pole and a force directed upwards.
MOTOR / GENERATOR
Successful application of flywheel energy storage
technology requires compact, reliable and highly efficient
motors/generators to transfer energy into and out of flywheels.
High-speed operation and high reliability requirements limit
selection of motors/generators to brushless and windinglessrotor types. Among these, permanent magnet (PM) machines
have the most advantages, including higher efficiency and
smaller size when compared with other types of
motors/generators of the same power rating, such as induction
or variable reluctance machines. They also exhibit lower rotor
losses and lower winding inductances, which make them more
suitable for a vacuum operating environment and the rapid
energy transfer in flywheel applications. Very low cogging
torque and robust rotor construction with very low part count
are additional arguments for using PM motor/generators in
flywheel applications.
Due to its robust rotor construction, a PM machine is well
suited for high-speed operation. High speeds allow one to
obtain the required power output with a small torque, making
the machine power dense and compact. To illustrate this point,
Figure 7 shows a comparison of a 2MW PM generator of
similar topology that was developed at Calnetix for a marine
application, side by side with a conventional low speed
machine of the same rating.
Rotor laminations
Radial control
windings
Solid iron
stator poles
73”
(1.85m)
high
28”
(0.71m)
high
Stator
laminations
Bias magnet
Axial control
winding
Figure 6. Axial section of the combo magnetic
bearing and explanation of the axial force
generation. Red (continuous) lines indicate bias
flux, blue (dashed) – axial control flux.
The new design, similar to the 125kW flywheel, will
accommodate mechanical backup bearings to support the rotor
in the event of magnetic bearings failure or overload. Since
active magnetic bearings continuously monitor rotor position,
some of the rotor failures can be detected in advance and the
system shutdown before a serious damage occurs. This feature,
along with large design safety margins and adequate
containment, results in a very safe operation.
Figure 7. Dimensional comparison between
Calnetix existing high-speed PM generator (in front
in color) and conventional 2 MW machine (in the
back).
With comparable power electronics that is designed for pulsed
mode of operation (to be discussed in more details in the next
section), the 2MW generator shown in Figure 7 would be
capable of a short output pulse of at least 6 MW or 3 times its
continuous rating. This is typical for PM machines where a
machine’s pulsed output rating is 2 to 5 times its continuous
power rating, with the latter being mainly limited by thermal
issues. Rotor overheating is a particularly dangerous thermal
issue for flywheels as the rotor is suspended without a
mechanical contact and cooling possibilities are limited to
radiation.
6
Copyright © 2006 by ASME
POWER ELECTRONICS
A control strategy that was found to be very effective for
producing a high power on the output of a PMSM machine is
maintaining a torque angle δ, which is the angle between the
current vector and the rotor field axis at 900 in vector control.
This advanced control strategy can be implemented even with
the rapidly changing speeds of the FESS application [17].
A back-to-back three-phase IGBT-based (Insulated-Gate
Bipolar Transistor) FESS power electronics topology is shown
in Figure 8. In the motoring mode, when the flywheel is
charged, the front-end converter operates as a rectifier and the
machine-side converter operates as an inverter. In the discharge
mode inversely, the machine-side converter works as a
controlled rectifier while the utility side converter works as an
inverter.
TAp
vA
vB
vC
iA
Lr
TBp
A more detailed analysis of the armature reaction and
saturation effects revealed that in fact even higher output power
TCp
TRp
uA
iB
TSp
TTp
uR
iR
C
uS
uB
iC
uC
TAn
TBn
TCn
TSn
TRn
iS
uT
TTn
vRS
PMSM
vST
Flywheel
iT
Cr
Figure 8. Three-phase back-to-back IGBT-based FESS power electronics.
As mentioned earlier, the power output from a PM
machine depends not only on the machine itself but also on the
power electronics, defining the character of the load on the
machine terminals. With the same basic topology shown in
Figure 8, the power output from the generator can be changed
significantly by applying different control algorithms such as
constant torque angle control, optimum torque per ampere
control, unity power factor control, constant mutual air gap flux
linkages control, and flux weakening control. Compared to
purely resistive generator loading, a 2-5 times increase of the
machine’s power output can be expected. A possibility of
increasing power output of a PMSM (Permanent Magnet
Surface Mounted) machine in pulse mode using advanced
control algorithms has been investigated. To evaluate potential
advantages of different control strategies, the existing 0.64kWh
125kW 25-35kRPM FESS has been used as an example. The
effects of applying different control algorithms have been
evaluated through modeling the entire system in PSIM, a power
electronics and motor drive simulation software. The block
diagram of the flywheel control system is shown in Figure 9.
Vt
Terminal
OP
Voltage
Volts L-N rms
can be obtained with δ being somewhat different from 900.
Table 3 below shows output power of the existing FESS
calculated for different power factors and torque angles at
35,000 rpm. In case of the first operating point (OP1) the torque
angle has been set equal to 900 (Id = 0, Iq =579.5 Arms), while
in the second case (OP2), the torque angle is set at 700. As it
can be noticed, the power output in the second case is
significantly higher. The explanation for this phenomenon lies
in the effects of the armature reaction and saturation effects,
however, a detailed explanation is beyond the scope of this
paper. As can be learned from Table 3, utilizing advanced
control strategies allows power increase by a factor of three or
more compared to simple passive rectification featuring lagging
power factor, which can be close to unity in the best case
scenario. Very high powers can be produced only in a lowduty-cycle operation in order to avoid overheating the machine.
A part which is especially vulnerable to overheating is the rotor
since it is suspended in vacuum without any mechanical contact
and the heat extract is limited to radiation only.
Is
Phase Current
Amps-rms
Iq
Torque
Component
amps-rms
Id
Flux
Component
amps-rms
1
523.0
579.5
579.5
0.0
2
3
4
5
446.1
318.4
88.5
37.9
579.5
579.5
579.5
579.5
537.5
406.8
125.1
58.5
216.5
412.8
565.8
576.5
P.F.
Pout
Output Power Power Factor
kW
0.5132
466.6
(leading)
542.9
.7 (leading)
482.1
.871 (leading)
153.9
unity
62.6
.95 (lagging)
Table 3. FESS generator output power with different terminal load power factors and torque angles
at 35kRPM.
7
Copyright © 2006 by ASME
Machin-Side
Converter
T Rp TSp TTp
uR
ias
C
uS
v dsr
PI
+
Vbs
a, b ,c
PI
i dsr
uT
TTn
PMSM
vST
Flywheel
ics
resolver
TTn
TSn
TTp
TSp
TSn
vRS
Vas
d, q
v qsr
-
TRn
TRp
T Rn
ibs
Vcs
PWM
Generator
i qsr
d, q
+
i
*
i dsr
= 0
a b,c
-
ibs
θ
*
qsr
Current
ias
T ec
Speed
∗
PI
calculator
+
- ω
r
calculator
θ
Software
ω ref
Figure 9. Block diagram of FESS power electronics control system.
CONCLUSIONS
The use of commercial flywheel energy storage technology to
address the needs of modern electric navy ships has been
investigated. The advantages of FESS over commercially
available electrochemical batteries include:
1)
2)
3)
4)
5)
Much higher charge/discharge rate (output power)
Essentially unlimited cycle life
Observable state of charge
Minimal, if any, maintenance
Environmental friendliness
A FESS capable of producing 500kW×30sec in high-duty
mode and up to 2MW in pulse mode described in this paper
targets applications in power backups for onboard electrical
power network. This power network is being developed by the
Navy for a new generation of more electric ships. There is also
a very promising commercial application for this FESS in load
leveling and energy recovering at electrical train stations.
The proposed flywheel utilizes a hub made of conventional
high-strength steel, proprietary high-performance magnetic
bearings and a permanent magnet brushless motor/generator.
The steel is chosen over more modern composite materials in
order to avoid integration problems associated with the latter.
Instead of minimizing the rotor weight through using lowdensity high-strength composites, the large magnetic surface
area of a steel rotor is used to passively lift it using permanent
magnets.
Active magnetic bearings are used to stabilize an otherwise
unstable system, compensate for inaccuracies of the passive
rotor weight offset and react dynamic loadings. Since they do
not have to deal with static weight, they can offer larger
dynamic load capacities overcoming the disadvantage of having
a heavier rotor. Due to PM-biased homopolar construction, the
active magnetic bearings used in the design offer very high
efficiency and very low rotational losses. These features are
further enhanced by utilizing the passive rotor lift.
PMSM motor/generator used in this design is well suited
for high-speed operation. High-speeds allow one to obtain the
8
Copyright © 2006 by ASME
required power output with small torque, making the machine
extremely power dense and compact.
Advanced power electronics control algorithms developed
at Texas A&M University are demonstrated to increase power
outputs of Calnetix’s PMSM machines by several times. These
power electronics make the machines more versatile as higher
power can be delivered in low-duty operational regimes when
thermal limitations are not as severe as in high-duty
applications.
High power density, high efficiency and high reliability of
the proposed design make it very attractive for both Navy and
commercial applications.
ACKNOWLEDGMENTS
This work is performed under US Navy STTR contract
N00014-04-M-0272. The authors greatly appreciate the support
and insightful technical suggestions provided by the Program
Officer Lynn Petersen.
REFERENCES
[1] Harvey E., Kingsley J., Matthew S., 2002, “United
States Navy (USN) Integrated Power System (IPS) Gas Turbine
Generator Set Test Experience”, ASME paper no. GT-200230260.
[2] Newell, J. M., Young S. S., 2000, “Beyond Electric
Ship” (preprint), publication of the Institute of Marine
Engineers, London, UK.
[3] Vander Meer, J., et al., 2005, “Improved Ship Power
System – Generation, Distribution, and Fault Control for
Electric Propulsion and Ship Service”, Proceedings of the IEEE
Electric Ship Technologies Symposium, Philadelphia,
Pennsylvania, pp. 284-291.
[4] Hawkins, L. A. et al., 2005, “Development of an AMB
Energy Storage Flywheel for Commercial Application”,
Proceedings of the 8th International Symposium on Magnetic
Suspension Technology, Dresden, Germany, pp. 261-265.
[5] Hawkins, L. A. et al., 2003, “Development of an AMB
Energy Storage System for Industrial Applications”,
Proceedings of the 7th International Symposium on Magnetic
Suspension Technology, Fukoka, Japan, pp. 156-163.
[6] McMullen, P. T. et al., 2003, Design and Development
of a 100kW Energy Storage Flywheel for UPS and Power
Conditioning Applications, Proceedings of the 24th
International PCIM Conference, Nurenberg, Germany, pp. 213218.
[7] Hawkins, L. A. et al., 2002, “Shock and Vibration
Testing of an AMB Supported Energy Storage Flywheel”,
Proceedings of the 8th International Symposium on Magnetic
Bearings, Mito, Japan, pp. 581-586.
[8] Hawkins, L. A. et al, 2001, “Influence of Control
Strategy on Measured Actuator Power Consumption in an
Energy Storage Flywheel with Magnetic Bearings, Proceedings
of the 6th International Symposium on Magnetic Suspension
Technology, Turin, Italy, pp. 42-47.
[9] Hawkins, L. A. et al, 2000, “Analysis and Testing of a
Magnetic Bearing Energy Storage Flywheel with GainScheduled MIMO Control”, ASME paper no. 2000-GT-405.
[10] Hawkins, L. A. et al, 1999, “Application of Permanent
Magnet Bias Magnetic Bearings to an Energy Storage
Flywheel”, Proceedings of the 5th International Symposium on
Magnetic Suspension Technology, Santa Barbara, California,
pp. 309-324.
[11] McGroarty J., Schmeller J., Hockney R., Polimeno
M., 2005, “Flywheel Energy Storage System for Electric Start
and an All-Electric Ship”, Proceedings of the IEEE Electric
Ship Technologies Symposium, Philadelphia, Pennsylvania, pp.
400-406.
[12] www.torayca.com
[13] Genta, G., 1985, Kinetic Energy Storage, Butterworth
& Co. Ltd., London, UK, p.53.
[14] Earnshaw, S., 1842, “On the nature of the molecular
forces, which regulate the constitution of the luminiferous
ether”, Transactions of Cambridge Philosophical Society, vol.
7, pp. 97-112.
[15] Braunbek, W., 1939, “Freischwebende korperim
elektrischen und magnetischen feld,” Z. Phys., vol. 112, pp.
753–763.
[16] Braunbek, W., 1939, “Freies schweben
diamagnetischer korperim magnetfeld,” Z.Phys.,vol. 112, pp.
764–769.
[17] Toliyat H.A. et al., 2005, “Advanced High-Speed
Flywheel Energy Storage Systems for Pulsed Power
Applications”, Proceedings of the IEEE Electric Ship
Technologies Symposium, Philadelphia, Pennsylvania, pp. 379386.
9
Copyright © 2006 by ASME